1/14    Re: deep blue thread yesterday.  I actually met the designer of deep
        blue, Feng-Hsiung Hsu, a couple of times because he's a friend
        of my wife's family.  People just call him "deep blue".  :-)  After
        he left IBM, VCs started throwing money in his direction for his
        startup.  Not sure what happened to his company.  -taiwanese guy
        \_ my summary of this thread:
           <DEAD>www.ucomics.com/foxtrot/2004/01/13<DEAD>
           \_ Luddite!
        \_ I've read up on some URLs.  The Deep Blue team was allowed to
           modify the program between games in the tournament.  These
           modifications were made with the help of human grandmasters.
           Deep Blue didn't beat Kasparov.  Rather, Deep Blue with between-
           game tweaking by human grandmasters beat Kasparov.  That closes
           the issue in my book.  (That Kasparov went nuts and was incorrect
           in suggesting that Deep Blue was being modified during the game
           doesn't change the main point.)
           \_ Yes, but Kasparov consults (quite frequently) with fellow
              grandmasters during (when recessed) and between games. He also
              employs computers to look for variations. So the resources
              available to the Deep Blue team are available to Kasparov.
              Similarly between games Fischer would consult others for advise
              and modify his tactics according to past games. So it's not really
              fair to say that the Deep Blue team can't modify the program
              to reflect what essentially amounts to learning. A human tweaks
              himself between games, so why can't a computer program be
              tweaked in between games?
              \_ wait a second. he talks to grandmasters *during* games? how can
                 that be? that's like chess by committee. I think you're wrong
                 there. Or else grandmasters advising Deep Blue would be ok.
              \_ You could even argue that Deep Blue was "talking with
                 other grandmasters and improving its strategy" between games,
                 just like Kasparov could do the same.  But while it
                 sounds appealing, I'm kind of uneasy about this reasoning.
              \_ I think it would be better if neither side could do any out
                 side consulting or tweaking.
                 My gut feeling was that Kasparov was alone here, and
                 the corporate resources of IBM way outmatched him in the
                 between-game improvements.
           \_ Well in the end the machine played by itself each game. Being
              tweaked before and after games doesn't really change that. GK
              made some mistakes too. It's not like they would know what
              Kasparov would do in the next game.
              \_ Some would argue that the engine was so tweaked to play
                 Kasparov, that it wasn't a general engine, but rather a "beat
                 Kasparov" engine.  We'll never know, since it didn'y play
                 anyone else.
                 \_ That's true for any world-class chess player preparing
           the issue in my book.
                    for a match.  Such a player would spend a lot of time
                    optimizing his play for his opponent by studying his games,
                    preparing tailored opening surprises, etc.  -- ilyas